Safety climate and safety culture in health care and the petroleum industry : psychometric quality, longitudinal change, and structural models

Abstract

This thesis presents four empirical studies of safety climate and one study of a safety culture programme. Four aims were defined to guide the conducted work: 1) assess safety climate in health care and petroleum and evaluate the psychometric properties of instruments used; 2) study the stability versus changeability of safety climate over time; 3) investigate the possibility of identifying a common safety climate structural model in health care and petroleum; and 4) improve knowledge about the dynamics and effects of safety culture programmes through development and testing of a structural model. Longitudinal designs were used to assess safety climate in both specialised health care and among workers on offshore platforms over a two-year period. In addition, cross-sectional data were explored to investigate the possibility of a common safety climate model in specialised health care and petroleum. A mixed method design was used to develop and validate a structural model that could illustrate the dynamics of a safety culture programme implemented in a large petroleum company. Hospital Survey on Patient Safety Culture (HSOPSC), developed by Sorra and Nieva (2004), was translated into Norwegian and used to assess safety climate at a large regional hospital in two measurement waves. A new instrument—Norwegian Offshore Risk and Safety Climate Inventory (NORSCI)—was similarly used to assess safety climate on the Norwegian continental shelf during two measurement waves. Two-year intervals were used between measures. The psychometric properties of the Norwegian version of HSOPSC and NORSCI were considered valid and satisfactory for both measurements. Results from the hospital suggest that the safety climate level was relatively stable during the measurement period, indicating that implemented improvement efforts have had relatively little impact on safety climate dimensions. Three safety climate dimensions were improved, two were reduced, and five had no significant change. On the continental shelf, four safety climate scores improved during the period, while one dimension was lowered.After adaptations, development, and validations of comparable safety climate concepts based on HSOPSC, factor analysis revealed six identical cross-industrial measurement concepts that could be used to develop a common structural model in health care and petroleum: 1) learning, feedback, and improvement; 2) teamwork within units; 3) supervisor/manager expectations and actions promoting safety; 4) transitions and teamwork across units; 5) organisational management support for safety; and 6) stop working in dangerous situations. A structural model assessment supported the hypothesised structural links specified among these dimensions—namely, safety climate variables at higher organisational levels influenced safety behaviour via safety climate dimensions at lower organisational levels. However, this model could only be generalised to fit health care and offshore petroleum workers, not onshore petroleum workers. To explore the dynamics of a safety culture programme, a mixed method approach was adopted using qualitative interviews, fieldwork, and a questionnaire survey. Five measurement concepts were validated and incorporated into a hypothetical structural model: 1) participation in a two-day kickoff; 2) effectiveness of programme implementation; 3) personal programme commitment; 4) safety behaviour change; and 5) safety culture change. The final model developed illustrates in particular how the levels of personal programme commitment and effectiveness of programme implementation influence the level of change regarding safety behaviour and safety culture. Another finding was that participation in the two-day kickoff had both positive and negative influences due to the high expectations developed among workers. Safety behaviour change influenced safety culture change and vice versa, which was also hypothesised.